Short: Memory Leaks in Darkwind Date: Mon, 04 Mar 2002 12:41:46 -0500 From: Scott Anderson <scott@chapman-anderson.org> Type: Bug State: New See also: ferrycode/ subdirectory Is this what you're looking for? I trimmed the log considerably... this is representative of what it contains. Thanks, -scott Lars Duening wrote: >Scott Anderson wrote on Sunday, March 03 2002, 10:08:00: > >>How can the total from status be 3 times the malloc? :-) >> > >There is an accounting error with the arrays and mappings - I don't quite >know where it is though. In the 3.2.9 line I added some diagnostic code for >this, seems that I need to add it to 3.3 as well. > >>GC seems to work well... the free small blocks size (g) went through the >>roof. >> > >A bit too well - the driver must have leaked a lot of memory. Did you use >MALLOC_TRACE and MALLOC_LPC_TRACE when configuring the driver? If yes, it >would be nice if you could send me a copy of the GC log so that I can take a >closer look at the leaks. > ------------------------- tabled string 08343d34 '/room/' was left unreferenced, freeing now. tabled string 083448ac '/obj/' was left unreferenced, freeing now. tabled string 08343ca8 '/secure/include/' was left unreferenced, freeing now. scanning chunk 08b57b20, end 08b5bb40 for unref'd blocks scanning chunk 08b53a20, end 08b57a40 for unref'd blocks scanning chunk 08b4f920, end 08b53940 for unref'd blocks freeing small block 0x08b50010 (user 0x08b5002c) interpret.c::allocate_uninit_array 13163 By object: objects/ships/ferry/maru#24 By program: Not found at old address. -- All leaked blocks in ferries at this line. scanning chunk 082f23e0, end 082f6400 for unref'd blocks scanning chunk 082ee2e0, end 082f2300 for unref'd blocks scanning chunk 082ea1e0, end 082ee200 for unref'd blocks scanning chunk 081a2ae0, end 081a6b00 for unref'd blocks 509 small blocks freed 2002.03.04 09:22:00 Consolidation merged 3640 blocks, freed 0 chunks holding 0 blocks. 2002.03.04 09:22:00 GC freed 0 destructed objects. ------------------------------------------------------- Date: Tue, 05 Mar 2002 14:16:15 -0500 From: Scott Anderson <scott@chapman-anderson.org> >Since all these allocations happen in the same, otherwise >unremarkable spot, I wonder if this might be a problem with the >mudlib code (self-referential array or the like). > >Can you give me a copy of the ferry code? > This should have all the relevant files. Also, if you want, log on (darkwind.org 3000) and I'll make you a wizard. We've got two muds, dev and live, and if it helps you to debug things you can recompile the driver on dev and bounce it pretty much as you wish.
